Introduction

Brain-derived neurotrophic factor (BDNF) is a neurotrophin family member mainly expressed in the brain, in both the hippocampus and cortex, but also present at the peripheral level1. BDNF is responsible for the development and the structural and functional maturation of key brain areas, with roles in branching, synaptic differentiation, pruning and plasticity1,2,3,4,5. BDNF synthesis, like that of other neurotrophins, occurs in the endoplasmic reticulum, where it is produced as an inactive precursor, pre-pro-BDNF. Its pre-region is then cleaved in the Golgi apparatus to form the immature but biologically active pro-BDNF isoform (28–32 kDa)6,7, which can be further excised, both intra- and extracellularly, by specific proteases into mature BDNF (mBDNF; 14 kDa), also biologically active6,7. Both isoforms exhibit a “yin-yang” effect, eliciting opposite biological outcomes8. Specifically, pro-BDNF binds to the neurotrophin receptor p75NTR, triggering apoptotic signalling and neuronal pruning1,7, while mBDNF selectively binds to tropomyosin kinase B (TrkB) receptor, promoting neuronal survival and growth4,9. Given its neurotrophic properties, mature BDNF is essential for neuronal plasticity and learning1,4,7,10.

BDNF is a promising clinical effect biomarker, measurable non-invasively in several biological matrices11 and at various levels of biological complexity12,13,14. Numerous studies have shown that changes in BDNF levels are associated with a wide range of pathologies, including neuropsychiatric disorders and neurodegenerative diseases15,16,17,18,19,20, as well as non-neurologic conditions21,22,23. It should be noted that BDNF is not only of clinical relevance but also relevant in Human Biomonitoring Programs10,24. In this context, such effect biomarkers are recognized as an ideal tool for investigating the relationships between exposure to environmental chemical and adverse health outcomes in population-based studies, as well as for elucidating some of the mechanisms underlying these associations6,10,24. In fact, recent epidemiological evidence has revealed a promising relationship between exposure to bisphenol A (BPA), non-persistent pesticides and metals with circulating levels of total BDNF protein12,13,14. However, some contradictions exist, suggesting that the pro- and mature-BDNF isoforms could help to clarify these inconsistencies12,13,14. In line with this hypothesis, recent studies suggest that the BDNF/pro-BDNF ratio is altered in neurological diseases, leading to an imbalance in serum levels of pro- and mBDNF, which may play an important role as biomarkers in these pathologies8,25,26,27. If pro- and mBDNF are indeed functionally related, it would be expected that their peripheral concentrations correlate with brain functionality and covary over time. These findings highlight the importance of incorporating these BDNF isoforms as effect biomarkers in human clinical and biomonitoring studies, aiming to enhance our understanding of the public health implications of human exposure to environmental chemicals.

Although the secretion of pro-BDNF by central nervous system (CNS) neurons has been debated28, recent studies have explored the relevance of BDNF isoforms as biomarkers in non-invasive body fluids, using validated immunoassays29. Similarly to total BDNF protein levels, the most common method for quantifying protein levels of pro- and mBDNF is enzyme-linked immunosorbent assay (ELISA), supporting their measurement based on previous findings30,31,32. ELISA, based on antigen–antibody reaction, is a straightforward, inexpensive, and cost-effective technique that enables the efficient and high-throughput analysis of a large number of samples with high specificity and sensitivity30,33. However, previous studies have indicated that some antibodies used in ELISA kits available on the market for quantifying pro- or mBDNF may exhibit low specificity, often failing to distinguish between both isoforms31,32. Thus, the aim of the present study was to identify the most commonly used commercial serum pro- and mBDNF assays in the literature, and to compare their performance in terms of specificity, accuracy, sensitivity and reproducibility in human serum samples.

Results

Of the 27 articles selected in the search (Supplementary Figure S1), 2 studies quantified total BDNF, 22 quantified pro-BDNF and 24 quantified mature BDNF in human serum samples. The most common brands used were Aviscera-Biosciencie (n = 10) and R&D Systems (n = 9), followed by FineTest and Biosensis (n = 3). Minority brands included Enzyme-linked Biotechnology (n = 2), Promega (n = 2) and Millipore (n = 1). Interestingly, one study reported using an ELISA developed by the authors themselves34, while 2 articles did not provide information on the exact ELISA kit used16,35. The main information about those is summarized in Supplementary Table S1.

Comparison of ELISA kits

Human serum samples were analysed for both isoforms (pro-BDNF and BDNF) in three different brands and at three dilutions per assay, chosen according to: (i) the manufacturer’s recommendations (Table 1), (ii) the scientific literature (Supplementary Table S1), and (iii) considering one common dilution for all commercial brands tested. Among the evaluated kits, the pro-Human BDNF DuoSet ELISA (#DY3175, R&D Systems) and the high-sensitivity pro-BDNF human ELISA (#SK00752-09, Aviscera Bioscience) did not provide specific recommendations for serum sample dilution. Therefore, a spiking/recovery test was performed to determine the optimal dilution factor. A 1:30 dilution yielded a recovery rate of 82.19% for the R&D Systems kit, while a 1:20 dilution resulted in a recovery of 96.80% for the Aviscera Bioscience kit (Supplementary Tables S2 and S3). Both values fall within the acceptable recovery range of 80–120%36. Table 2 summarizes the results obtained for total BDNF, pro-BDNF and BDNF serum concentrations in each of the ELISA kits tested.

Table 1 Summary of ELISA’s characteristics. Main characteristics declared by the manufacturers of each ELISA kit evaluated in this study.
Full size table
Table 2 ELISA kits tested. Comparison between the performance declared by the manufacturer and the one obtained in this study for each ELISA kit tested.
Full size table

The optical density (OD) values for the standards of all immunoassays tested were within the range established by the manufacturers, except for the last point of the standard curve (the most concentrated), which was higher than the declared in the human free BDNF Quantikine ELISA (#DBD00, R&D Systems), the high sensitivity pro-BDNF human ELISA kit (#SK00752-09, Aviscera Bioscience), the human pro-BDNF ELISA (#EH4255, FineTest) and the human BDNF ELISA kit (#EH0043, FineTest) (Supplementary Figures S2, S3 and S4). The OD values of the standards were not specified in the characteristics of the Human pro-BDNF DuoSet ELISA (#DY3175, R&D Systems).

Overall, the blank values obtained in the ELISA kits from Aviscera Bioscience (#SK00752-09, #SK00752-01) were 2-fold higher than those declared by the manufacturer. The total BDNF Quantikine ELISA (#DBNT00, R&D Systems) showed values 4-fold higher, while the ELISA kits from FineTest (#EH4255, #EH0043) and the human free BDNF Quantikine ELISA (#DBD00, R&D Systems) exhibited values 0.4-fold higher than those declared. However, the 4-PL standard curves obtained in all the ELISA kits showed an R2 value between 0.99 and 1 (Supplementary Figures S2, S3 and S4), and the results obtained showed the same range and sensitivity as those declared by the manufacturers (Supplementary Figures S2, S3 and S4).

Most tested conditions fit the linear region of the standard curve, but for BDNF, a few samples diluted at 1:20 and 1:40 in the human free BDNF Quantikine ELISA kit (#DBD00, R&D Systems) and in the high-sensitivity BDNF ELISA kit (#SK00752-01, Aviscera Bioscience), respectively, were above the highest declared point. Regarding pro-BDNF, all the conditions tested in pro-BDNF ELISA kits (#DY3175, R&D Systems; #SK00752-09, Aviscera Bioscience), as well as the 1:80 dilution in the human free BDNF Quantikine ELISA kit (#DBD00, R&D Systems), were found in the lower end of the standard curve.

Protein levels of pro-BDNF and BDNF in the 23 human serum samples showed an approximately normal distribution in all the commercial brands assessed, except for all dilutions tested in the high-sensitivity pro-BDNF human ELISA kit (#SK00752-09, Aviscera Bioscience) and the 1:100 dilution in the human pro-BDNF ELISA (#EH4255, FineTest). Pro-BDNF levels were always lower than BDNF concentrations at the different dilutions and in all kits tested (Fig. 1; Table 2). Median and percentile values (25th and 75th) for pro-BDNF were 6.57 (5.69–6.88) ng/mL in the human pro-BDNF DuoSet ELISA kit (#DY3175, R&D Systems) for serum samples at the optimal dilution of 1:30, 9.94 (7.93–11.20) and 7.87 (5.93–10.25) ng/mL at the 1:100 and 1:200 optimal dilutions in the human pro-BDNF ELISA kit (#EH4255, FineTest), and 8.80 (8.14–9.01) ng/mL in the high sensitive pro-BDNF ELISA kit (#SK00752-09, Aviscera Bioscience) (Table 2). BDNF protein levels were 32.16 (23.47–36.59) ng/mL at an optimal dilution of 1:20 in the human free BDNF Quantikine ELISA (#DBD00, R&D Systems) (Table 2). Figure 2 presents the Spearman correlations among BDNF levels measured in the set of 23 human serum samples. In R&D systems kits, total BDNF (1:100 dilution) was positively and significantly correlated with BDNF at the 1:20 dilution (rho = 0.94, p < 0.001), but not with pro-BDNF at the 1:30 dilution (rho = 0.06, p = 0.771). R&D systems pro- and BDNF levels did not correlate (rho = 0.08, p = 0.707). Additionally, total BDNF strongly correlated with the sum of pro-BDNF at the 1:30 dilution and BDNF levels at the 1:20 dilution (rho = 0.94, p < 0.001) in R&D Systems. Regarding Aviscera Bioscience, pro-BDNF levels at the 1:20 dilution correlated significantly with BDNF levels at the 1:160 dilution (rho = 0.48, p = 0.02; Fig. 2). In the FineTest kits, pro-BDNF levels at the 1:100 dilution correlated significantly with both tested dilutions of BDNF (1:100, rho = 0.84, p < 0.001; 1:200, rho = 0.63, p < 0.001), while pro-BDNF levels at the 1:200 dilution were not correlated with BDNF levels (1:100, rho = 0.02, p = 0.922; 1:200, rho=−0.12, p = 0.603) (Fig. 2). Intra-assay and inter-assay CVs were assessed by comparing the protein values obtained in duplicate within the same plate and between different plates for each of the 23 serum samples, respectively (Supplementary Figure S5). As reported in Table 2, all kits used to quantify pro- and BDNF presented valid values for both CVs, and in all cases, lower than those declared by the manufacturers (the R&D Systems pro-BDNF DuoSet ELISA kit #DY3175 did not declare CV values).

Fig. 1
figure 1

Levels of proteins (ng/mL) obtained in human serum samples (n = 23). Different dilutions were assessed in ELISA kits from R&D Systems (A, B and E) [total BDNF Quantikine ELISA kit#DBNT00 (A), Human pro-BDNF DuoSet ELISA kit #DY3175 (B) and human free BDNF Quantikine ELISA kit #DBD00 (E)], Aviscera Bioscience (C and F) [high sensitivity pro-BDNF human ELISA kit #SK00752-09 (C) and high sensitive BDNF ELISA kit #SK00752-01 (F)] and Finetest (D and G) [human pro-BDNF ELISA kit #EH4255 (D) and human BDNF ELISA kit #EH0043 (G)]. Comparison of protein levels obtained in human serum samples diluted at 1:20 (H) for pro-BDNF and at 1:40 (I) for BDNF (common dilutions) in the ELISA kits from the 3 companies used in this study (R&D Systems, Aviscera Bioscience and FineTest). Each value obtained of the samples analysed was represented, and the median ± SD of all of them.

Full size image
Fig. 2
figure 2

Correlation’s heatmaps. Results about the Spearman correlations obtained among serum protein levels (total, pro-BDNF and BDNF) obtained in the different commercial brands used in this study [R&D Systems, Aviscera Bioscience and FineTest].* p-value < 0.05.

Full size image

Regarding total processing time, all ELISA kits measuring BDNF (#DBD00, #SK00752-01 and #EH0043) and the FineTest´s pro-BDNF kit (# EH4255) required less time (~ 4 h) than the Aviscera Bioscience (#SK00752-09) and R&D Systems (#DY3175) pro-BDNF kits, which needed 6’5 h and overnight incubation plus additional time of 7 h, respectively (Table 2).

Qualitative and quantitative confirmatory specificity analysis of BDNF isoforms

The Western blot assay (qualitative analysis) was performed using the detection antibodies provided by each kit for pro-, mature BDNF, or both (total BDNF), to verify the BDNF data obtained by the ELISA kits.

A negative control (BSA), together with commercially available human pro- and mature BDNF, were used to identify the reactivity of antibodies from the commercial kit brands tested. The Western blot results showed that antibodies from the R&D Systems total BDNF Quantikine ELISA kit (#DBNT00) reacted with both pro- and mature BDNF isoforms (Fig. 3A). Antibodies from the R&D Systems human pro-BDNF DuoSet ELISA kit (#DY3175), the Aviscera Bioscience high sensitivity pro-BDNF human ELISA kit (#SK00752-09) and the FineTest human pro-BDNF ELISA kit (#EH4255) reacted specifically with pro-BDNF; although minimal cross-reactivity with mature BDNF was found in all of them (Figs. 3B, D and F). The R&D Systems human free BDNF Quantikine ELISA kit (#DBD00) was marketed for mature BDNF, showing specificity but also marginal cross-reactivity with pro-BDNF (Fig. 3C). In contrast, the Aviscera Bioscience high-sensitivity BDNF ELISA kit (#SK00752-01) and the FineTest human BDNF ELISA kit (#EH0043), also marketed for measuring mature BDNF, showed a higher reactivity to pro- than to mature BDNF (Figs. 3E and G).

Fig. 3
figure 3

Western blot for qualitative analysis of pro-BDNF and mBDNF antibodies. Antibodies from R&D Systems [total BDNF Quantikine #DBNT00 (A), human pro-BDNF DuoSet #DY3175 (B) and human free BDNF Quantikine #DBD00 (C)], FineTest [human pro-BDNF #EH4255 (D) and human BDNF #EH0043 (E)], and Aviscera Bioscience [high sensitivity pro-BDNF human #SK00752-09 (F) and high sensitive BDNF #SK00752-01 (G)] ELISA kit were tested. Original blots are presented in Supplementary Figure S6.

Full size image

To complement the Western Blot results, the percentage of cross-reactivity (quantitative analysis) was also evaluated for all ELISA kits used. Commercially available pro-BDNF and mature BDNF proteins were used to determine the cross-reactivity values. Table 2 shows that the pro-BDNF kits from the three brands tested showed negligible cross-reactivity with mBDNF: 1.3% (#DY3175, R& Systems), 1.5% (#EH4255, FineTest) and 1.7% (#SK00752-09, Aviscera Bioscience). The human free BDNF Quantikine ELISA kit from R&D Systems (#DBD00) showed a 9.4% cross-reactivity with pro-BDNF (Table 2). In contrast, the BDNF kits from Aviscera Bioscience (#SK00752-01) and FineTest (#EH0043), marketed for quantifying mature BDNF, were more specific for pro-BDNF, as mentioned above (Figs. 3E and G). When cross-reactivity with mBDNF was tested, percentages of 24.3% and 15.3% were found, respectively, suggesting that these kits measure both pro-BDNF and mature BDNF to varying degrees and are therefore not specific (Table 2).

Discussion

The purpose of this study was to establish the most appropriate methodology to quantify pro- and mBDNF in human serum samples, identifying the best ELISA kits among the available commercial brands, based on their specificity, sensitivity, accuracy and reproducibility for quantifying protein levels for both isoforms. Our results showed that the total BDNF Quantikine ELISA kit (#DBNT00) from R&D Systems and the different pro-BDNF ELISA kits from the three brands tested [Human pro-BDNF DuoSet ELISA (#DY3175, R&D Systems), human pro-BDNF ELISA (#EH4255, FineTest) and high sensitivity pro-BDNF human ELISA (#SK00752-09, Aviscera Bioscience)] are the most suitable candidates. Although none of the ELISA kits marketed for mBDNF showed complete specificity for this isoform, the human free BDNF Quantikine ELISA kit (#DBD00, R&D Systems) was the unique assay that measured mature BDNF with an acceptable specificity and may serve as a suitable option for estimating serum mBDNF levels.

The available scientific information in relation to the quantification of BDNF isoforms in human samples is scarce, and the studies have primarily focused on populations with neurological problems due to the added value of BDNF as a biomarker in these pathologies (Supplementary Table S1). The evidence shows great heterogeneity in BDNF isoform levels in the population (Supplementary Table S1), probably due to genetic polymorphisms37, gender38, lifestyle (diet, physical activity, sleep patterns and alcohol intake)39,40,41 and technical factors (fasting vs. non-fasting and time of day)38,42,43,44. Based on our results, variations in kit performance may also be a relevant contributor to this heterogeneity. Therefore, it is necessary to develop standardised operational protocols to obtain comparable and homogeneous data among different scientific studies.

Total BDNF, and its isoforms, are involved in brain development, both in the process of neurogenesis and neuroplasticity, from infancy to adolescence1,6,45. Specifically, pro- and mBDNF are key regulators of spine pruning and maturation, respectively, and both are required for the refinement of neuronal connections in brain development5. However, despite the role played by pro- and mBDNF in different brain processes, there is currently no evidence on the levels of these isoforms in healthy populations, including children and adolescents. Concerning pro- and mBDNF, a balance between both isoforms has been described, which may be affected in adult neuropsychiatric or neurodegenerative disorders8. For example, in vivo studies have shown that inhibition of the pro-BDNF to mBDNF conversion leads to autism spectrum disorder (ASD)-like phenotypes27, suggesting that the pro-BDNF/BDNF ratio could serve as a potential biomarker for this disorder. Hence, there is a lot of interest in quantifying BDNF isoforms through a validated and reliable methodology, in order to use them as clinical and epidemiological biomarkers. In our work, the BDNF protein levels obtained were in line with those described in the literature; as expected, the levels of total BDNF were higher than those of its isoforms, and in turn, the levels of mBDNF were higher than those of pro-BDNF (Table 2 and Supplementary Table S1). In all tested conditions, variability in total, pro- and BDNF level was observed across the 23 human serum samples, except for pro-BDNF levels using the high-sensitive pro-BDNF ELISA kit (#SK00752-09, Aviscera Bioscience) (Table 2).

The R&D Systems brand offered, a priori, the opportunity to quantify the protein levels of total BDNF and both isoforms, and to study the relationships between them. It is important to note that, of the three brands that marketed kits claiming to measure mature BDNF, only the R&D Systems brand (#DBD00) was specific for mature BDNF in agreement with Polacchini et al31. Nevertheless, this kit (#DBD00) showed 9.4% cross-reactivity with pro-BDNF, so efforts should be made in the future to reach the < 5% threshold commonly established in ELISA guidelines46. Regarding, the R&D Systems pro-BDNF kit (#DY3175), it only showed 1.3% of cross-reactivity with mBDNF, being considered highly specific (Fig. 3B).

Specificity is a key element in distinguishing between pro-BDNF and mBDNF isoforms, which show opposite roles in brain functioning30,33. For this reason, we performed both a qualitative analysis of the antigen size through Western Blot47 and a quantitative analysis of antigen levels through the cross-reactivity test33. The three ELISA kits selected for pro-BDNF quantification in this study showed good specificity and acceptable percentages of cross-reactivity with mBDNF: 1.3% (#DY3175, R& Systems), 1.5% (#EH4255, FineTest) and 1.7% (#SK00752-09, Aviscera Bioscience) (Figs. 3B, D, F, and Table 2). However, the opposite occurred with the two ELISA kits that were supposed to measure mature BDNF [Aviscera Bioscience (#SK00752-01) and FineTest (#EH4255)], but which actually measured both isoforms (Figs. 3C, E, and G). The accuracy of the ELISA assays was determined by the linearity of the standard curve30,48. In all the kits used, accuracy was high, indicating a high degree of precision and reliability (Supplementary Figures S2, S3 and S4). Similarly, all ELISA kits tested showed high reproducibility, with intra- and inter-CVs lower than declared by the manufacturers (Table 2). Our results show that, unexpectedly, as the dilution increased, so did the pro-BDNF levels30,33 (Table 2; Fig. 1). This could be the result of a matrix-related effect, as serum components may interfere by reducing the recovery of the target analyte (Supplementary Tables S2 and S3). A high sensitivity is another key aspect of bioassays, as it allows the detection of low levels of the antigen in the biological fluid30. Our results showed that total BDNF and its isoforms were quantified in 100% of the samples analyzed at different dilutions, showing an acceptable sensitivity.

Our results are in line with previous reports emphasising that the available ELISA kits to quantify BDNF isoforms, especially mBDNF, need to increase their specificity to be suitable as biomarkers in biomonitoring programmes as well as in clinical applications31,32. The results obtained with the ELISA kits for mature BDNF [#DBD00 (R&D Systems), #EH0043 (FineTest) and #SK00752-01 (Aviscera Bioscience)] indicate that the antibodies used do not exhibit absolute specificity for this isoform. Among them, the R&D Systems mBDNF ELISA kit (#DBD00) could be a good estimator of serum mBDNF levels, as it showed a relatively high specificity, although with a 9.4% cross-reactivity with pro-BDNF. Indeed, the mBDNF isoform is technically difficult to measure, as antibodies directed against mature BDNF also recognise pro-BDNF, since the pro-BDNF isoform contains the mature BDNF sequence within its structure31,32. This cross-reactivity represents a significant limitation hindering the accurate and specific quantification of mBDNF using immunoassays.

Based on the challenging measure of mBDNF, an alternative strategy, although not without limitations, could involve measuring total and pro-BDNF levels separately using valid and specific commercial ELISA kits, and then estimating the concentration of mBDNF by calculating the difference (total BDNF minus pro-BDNF levels). This strategy was applied to the 23 serum samples as a post-hoc hypothesis. Using ELISA kits from the same manufacturer (R&D Systems), the comparison between the measured and estimated mBDNF showed a seemingly acceptable variation of ± 15% (Supplementary Table S4). In contrast, when kits from different manufacturers were combined, the variation increased to ±19% (total BDNF from R&D Systems and pro-BDNF from FineTest; Supplementary Table S5) and ± 21% (total BDNF from R&D Systems and pro-BDNF from Aviscera Bioscience; Supplementary Table S6). This would support a more reliable strategy when both measurements are performed using ELISA kits from the same manufacturer, which would guarantee greater specificity, sensitivity, accuracy, and reproducibility.

The development of ELISA kits to adequately measure pro- and mBDNF levels in different biological matrices is rapidly expanding. This includes the recent Fujifilm kits for measuring mature BDNF (Kit Wako, High Sensitive #290–85801 and #296–83201), in which manufacturers state a 1.3% and 10% of cross-reactivity with pro-BDNF, respectively, because the monoclonal antibody reacts with an N-terminal cutting surface of mBDNF, enhancing its specificity. The improved sensitivity of these kits could make them well-suited for other matrices besides serum, where a more sensitive detection limit is needed, such as human saliva49 and plasma50.

This study presents different strengths such as the use of several commercially available kits selected from a structured literature search, being representative of the current kits used by scientists in the field; the qualitative and quantitative assessment of the specificity of the antibodies in each brand tested; the protocol followed to reduce the variability of serum protein levels avoiding freeze-thaw cycles that favour the degradation of the protein; and the biological sample selected, serum instead plasma, as a matrix to measure human BDNF and its isoforms at higher levels45,51. Additionally, the ELISA kits that demonstrated the best specificity, sensitivity, precision, and reproducibility for serum samples would, in principle, be valid for analyzing human plasma samples. However, future works would be needed to identify the optimal plasma dilutions and to confirm adequate sensitivity, maintaining standardized collection and storage procedures until analysis. Several limitations should also be highlighted, such as the lack of knowledge of the age, sex and health status of the subjects included in this study; and the variability and heterogeneity shown by the protocols of the selected ELISA kits.

Conclusion

Our findings indicate that, among the commercial kits evaluated, the total BDNF Quantikine ELISA kit (#DBNT00) from R&D Systems and the three brands of pro-BDNF kits [Human pro-BDNF DuoSet ELISA (#DY3175, R&D Systems), Human pro-BDNF ELISA (#EH4255, FineTest) and High-sensitivity pro-BDNF human ELISA (#SK00752-09, Aviscera Bioscience)] would be the most reliable options for measuring serum levels of total BDNF and pro-BDNF, respectively. All of these immunoassays demonstrated high accuracy, sensitivity, reproducibility and specificity. However, none of the commercial kits evaluated for mBDNF provided optimal specificity due to antibody cross-reactivity. Nevertheless, the human free BDNF Quantikine ELISA kit (#DBD00, R&D Systems) showed the best specificity and may serve as a suitable option for estimating serum mBDNF levels. These results highlight the challenge in specifically measuring mature BDNF due to antibody cross-reactivity. As such, estimating mBDNF levels by subtracting pro-BDNF from total BDNF, using validated kits, could be another valid alternative strategy that should be further investigated. Given the critical roles that total BDNF and its pro- and mBDNF isoforms play in brain function, developing highly specific ELISA methods, and/or implementing indirect estimation strategies, remains essential for advancing biomarker research and achieving reliable BDNF isoform protein measurements in human samples.

Materials and methods

Literature search

A search algorithm was entered in the MEDLINE/Pubmed database up to March 2024 using the following terms in all fields: [(“proBDNF” OR “pro-BDNF” OR “mBDNF” OR “mature BDNF”) AND (“enzyme-linked immunosorbent assay” OR “ELISA”) AND “Serum”]. Studies were considered eligible if the following criteria were met: (1) human studies; (2) full-text articles; and (3) articles in English or Spanish. Articles without any information about pro- or mature BDNF, ELISA and human serum samples were excluded. The bibliography of the retrieved studies was also reviewed for possible relevant publications. From the selected articles, information was extracted on the subjects (sample size, health status, sex and age), the BDNF isoform analysed, the ELISA kit used (reference and brand), the conditions of analysis, the BDNF isoform concentrations obtained, as well as the authors and the year of publication.

Serum samples

Since the aim of this work was purely methodological, completely anonymized serum samples (n = 23) were collected from among those to be destroyed after completing the necessary analyses at the clinical analysis service of the Hospital Universitario Clínico San Cecilio (Granada, Spain). These samples had been collected in the routine practice of the hospital under all the necessary ethical requirements and guidelines of the aforementioned hospital, and informed consent was obtained from all participants. All methods were carried out in accordance to Hospital guidelines and Spanish regulations, and collection and informed consent protocols were approved by the ethics committee of the Hospital Universitario Clínico San Cecilio (Granada, Spain). The samples were centrifuged at 1500 g for 10 min at 4 °C. The resulting serum was aliquoted into 0.2 ml Eppendorf tubes and stored at −80 °C until further analysis at the Biomedical Research Center (CIBM) of the University of Granada (Spain).

Total, pro and BDNF measurements by ELISA

All serum samples were measured for total BDNF levels using the total BDNF Quantikine ELISA kit(Cat. #DBNT00, R&D Systems, Minneapolis, MN, USA) following our previously validated and applied protocol12,13,14,52. Building on our previous experience, both BDNF isoforms were quantified using R&D Systems brand to establish the relationship between total BDNF levels and its isoforms. The assays used included the human pro-BDNF DuoSet ELISA kit (Cat. #DY3175), and the human free BDNF Quantikine ELISA kit (Cat. #DBD00). Additionally, the two most commonly used ELISA kits for pro-BDNF referenced in the literature were also tested (Supplementary Table S1), namely, the high sensitivity pro-BDNF human ELISA kit (Cat. #SK00752-09, Aviscera Bioscience, Santa Clara, California, USA), and the human pro-BDNF ELISA kit (Cat. #EH4255, FineTest, Boulder, Colorado, USA). Moreover, the high sensitivity BDNF ELISA kit (Cat. #SK00752-01, Aviscera Bioscience), and the human BDNF ELISA kit (Cat. #EH0043, FineTest) were used to assess ‘mature’ BDNF levels. Biosensis, another brand, was not selected due to prior results by Polacchini et al.31, which reported that this ELISA kit (BDNF Rapid™ ELISA Kit, Cat. #BEK-2211) only quantified total BDNF and did not differentiate between isoforms.

The main characteristics of all ELISA kits used are summarized in Table 1. Spiking/recovery tests were performed to determine the optimal dilution factor, particularly for the kits lacking manufacturer-provided serum dilution guidelines [human pro-BDNF DuoSet ELISA kit (#DY3175, R&D Systems) and high-sensitivity pro-BDNF human ELISA kit (#SK00752-09, Aviscera Bioscience)]. All measurements were carried out at controlled room temperature, in duplicate, and by the same operator. Two ELISA plates with the same reference and lot number were used per assay to obtain intra-assay and inter-assay coefficients of variations (CV), an intrinsic indicator of the reproducibility of the kit (Supplementary Figure S5). Each protocol was executed in accordance with the manufacturer’s instructions. Optical density was measured using a microplate reader (MultiSkan, Fisher Scientific, United States), and a four-parameter logistic (4-PL) curve fit was performed to calculate the protein serum levels.

Qualitative and quantitative assessments of BDNF and its isoforms specificity

To confirm that each ELISA kit analyzed exactly the declared isoforms, we performed a qualitative assessment using Western blot assays. Samples were prepared with 16 ng of human - pro-BDNF (Cat. #B-256-B) from Alomone Labs (Jerusalem, Israel), mature BDNF (Cat. #orb1675341), from Biorbyt (Cambridge, UK), and Bovine Serum Albumin (BSA) as negative control (Cat. #A7030, Sigma-Aldrich, St. Louis, MO) mixed with 4X Laemmli sample buffer (Cat. #1610747, Bio-Rad Laboratories, CA, USA) containing 10% β-mercaptoethanol (Cat. #M3148, Sigma-Aldrich, St. Louis, MO) and milliQ-water. Samples were separated with SDS-PAGE using Mini-Protean TGX gel (Cat. #4561093, Bio-Rad Laboratories, CA, USA) and were transferred onto a nitrocellulose membrane (Cat. #1704159, Bio-Rad Laboratories, CA, USA). Membranes were incubated in blocking buffer [5% BSA in 1X Tris-buffered saline (TBS) with 0.05% Tween 20 (TBS-T)] for one hour to block the non-specific sites. The membranes were incubated with the primary antibodies provided in each ELISA kit at the concentrations (and diluents) suggested by the manufacturer overnight and washed 3 times in TBS-T (5 min per wash). As secondary detection of the membranes, streptavidin-horseradish peroxidase (HRP) provided in each ELISA kit was used at the concentration (and diluents) suggested by the manufacturer for one hour and then washed as described above. In the case of the total BDNF and human free BDNF Quantikine ELISA kits (R&D Systems; #DBNT00 and #DBD00, respectively), primary antibodies were HRP-conjugated, eliminating the need for further detection with a secondary antibody. Immunoreactive signals were detected using the Clarity Western ECL Substrate Kit from Bio-Rad (Cat. #1705061, Bio-Rad Laboratories, CA, USA). Membranes were digitally imaged with Image Reader ChemiDoc (Bio-Rad Laboratories).

Additionally, cross-reactivity with pro-BDNF and mBDNF isoforms was quantitatively assessed for all ELISA kits evaluated in this study using commercially available proteins: pro-BDNF (#B-256-B, Alomone Labs., Jerusalem, Israel) and mature BDNF (Cat. #SRP3014, Sigma-Aldrich, St. Louis, MO). Briefly, an assay was performed for each ELISA kit targeting the BDNF isoforms, in which two serum samples were loaded at the optimal dilution; and one of them was spiked with a known amount of pro-BDNF or mBDNF, depending on the isoform indicated by the ELISA manufacturer. The absorbance for each condition was measured, and the percentage of cross-reactivity was calculated using the following equation, adapted for non-competitive sandwich immunoassays53: Cross-reactivity (%) = [(OD of cross-reactant analyte)/(OD of target analyte)] ×100. All OD values were obtained at the same concentration, and the blank OD was subtracted. It is important to note that this equation is inverted relative to its original use in competitive ELISA50, to reflect the proportional relationship between OD and analyte concentration in sandwich formats.

Statistical analysis

Measures of central tendency and dispersion were calculated, including geometric means, medians, 25th and 75th percentiles, minimum and maximum for total, pro- and mature BDNF levels. The limit of detection (LOD) for each ELISA kit was defined as the lowest concentration of the standard curve11. The sensitivity was considered in line with declared by the kit, when the absorbance of the less concentrated standard was at least 10% higher than the absorbance of the blank. The accuracy was defined equal to the stated when the most concentrated standard point did not show a lower value R2 = 0.9. All samples were assessed in duplicates and intra- and inter-CV values calculated. Distribution of the BDNF levels (total and/or isoforms) was evaluated with the Shapiro-Wilk test. Spearman correlation test (rho) was performed to assess the relationship between serum protein levels (total BDNF, pro-BDNF and BDNF) obtained in the selected commercial brands. Statistical analyses were performed using SPSS version 28 (IBM SPSS, Armonk, NY, United States), with additional analyses carried out in Prism version 9.1.1 (GraphPad, California, United States) and Image J software.